What still needs validation?

Local Level Lab has two distinct projects. The primary project analyzes NASA airborne survey recordings. The secondary project asks whether stationary consumer IMUs recover a geographically consistent horizontal rotation field. An IMU (inertial measurement unit) measures turning and acceleration. Each project needs its own error model and validation.

NASA analysis

The completed archive comparisons and current numerical checks are described in the findings and analysis roadmap. Outstanding work includes solver qualification, calibration-boundary sensitivity, absolute model adequacy, timing/coordinate assumptions and dependence between recordings. These results do not certify consumer devices.

Validate observable predictions separately from their interpretation. The historical “rotating globe” label includes a specified background field; “still globe” omits it. A better fit does not identify the field's physical cause. The field follow-up must estimate its vector and magnitude freely before making rate comparisons. Neither a nominal-rate fit nor a latitude curve alone is a test of every cosmological explanation.

Stationary crowdsourcing

The app now collects still A/B blocks, original Bluetooth measurements, device information and participant-entered location/alignment. The preliminary analyzer fits a signed reversal rate without using latitude or an expected Earth rate. It does not call a recording a detection. Read the stationary method.

Before scientific pooling:

• Demonstrate preservation of independently imposed slow rotations and establish the practical digital resolution, warm-up behavior and repeatability of actual units.
• Freeze level, gap, duration and alignment rules from controls, independent of the measured rate and expected latitude curve. Preserve failed and null recordings.
• Measure nonlinear drift, temperature effects and temporal correlation. Calibrate uncertainty using controls and genuinely separate recordings, rather than treating every sample or block as independent.
• Test reversal errors and tilt, alternative starting orientations and known-null apparatus controls. Check whether errors repeat with instrument, support or operator.
• Seal per-recording estimates before comparing them with location. Fit the geographic amplitude freely and compare alternatives; do not calibrate to the expected answer.
• Replicate across physical instruments, days and sites. Preserve those groups when pooling. Two devices of one model are useful replication, not a guarantee of independence.

The initial 30-minute blocks are an adjustable development setting, not an established minimum. An overnight record is useful only if its error behavior is understood. This experiment does not target precise vertical rate, solar-versus-sidereal differences, polar wobble or small altitude effects.

Earlier aircraft development

Passenger-flight simulations and route searches remain preserved as historical work. They tested a different experiment. Their thresholds, usability domain and calibration certificates cannot be transferred to stationary recordings or NASA data.

The former public guide is retained in the aircraft validation history; the technical validation record retains detailed results. No historical campaign is rerun by this change.